Machine learning based call routing system

Abstract

Machine learning technology can analyze in real-time the data from a call between a person and a customer service representative. Based on this analysis, a server can determine a sentiment score that describes a sentiment expressed by the person or the customer service representative. If the server determines that the sentiment score is less than or equal to a pre-determined value, the server can inform the customer service representative's manager so that the manager can take further action to help the person and/or the customer service representative.

We claim: 
     
       1. A method comprising:
 receiving call data related to a conversation between two or more people including a first person and a second person on a call; 
 determining, using the call data while the first person is on the call with the second person, at least a first score associated with a first time on the call and a second score associated with a second time on the call; 
 determining a third score based on the first score and the second score, wherein the third score indicates a sentiment over time between the first time and the second time; and 
 in response to determining that the third score is less than or equal to a pre-determined score, storing the call in a queue. 
 
     
     
       2. The method of  claim 1 , further comprising:
 adjusting the third score based on a time of day when the call occurs, wherein adjusting the third score is performed by a machine learning model trained to adjust sentiment scores, wherein the machine learning model was trained using pairs of words to sentiment values. 
 
     
     
       3. The method of  claim 1 , further comprising:
 sending, to a third device associated with a third person, a selectable option to allow the third person to listen or speak on the call while the call is mapped to a second device of the second person. 
 
     
     
       4. The method of  claim 3 , further comprising:
 receiving, on a user interface of the third device, an indication that the third person selected the selectable option to join the call and, in response, routing the call to the third device of the third person while the call is connected to the second device. 
 
     
     
       5. The method of  claim 1 , further comprising:
 queuing the call by determining a number associated with a first device associated with the first person, wherein the number is used to initiate a second call. 
 
     
     
       6. The method of  claim 1 , wherein the first score and the second score are determined based on an inflection of a first voice of the first person or of a second voice of the second person. 
     
     
       7. The method of  claim 1 ,
 wherein the first time is within a pre-determined time window that begins from a start of the conversation on the call, and 
 wherein the second time is outside the pre-determined time window. 
 
     
     
       8. A system comprising:
 one or more processors; and 
 one or more memories storing instructions that, when executed by the one or more processors, cause the system to perform a process comprising:
 receiving call data related to a conversation between two or more people including a first person and a second person on a call; 
 determining, using the call data while the first person is on the call with the second person, at least a first score associated with a first time on the call and a second score associated with a second time on the call; 
 determining a third score based on the first score and the second score, wherein the third score indicates a sentiment over time between the first time and the second time; and 
 in response to determining that the third score is less than or equal to a pre-determined score, storing the call in a queue. 
 
 
     
     
       9. The system according to  claim 8 , wherein the process further comprises:
 adjusting the third score based on a time of day when the call occurs, wherein adjusting the third score is performed by a machine learning model trained to adjust sentiment scores, wherein the machine learning model was trained using pairs of words to sentiment values. 
 
     
     
       10. The system according to  claim 8 , wherein the process further comprises:
 sending, to a third device associated with a third person, a selectable option to allow the third person to listen or speak on the call while the call is mapped to a second device of the second person. 
 
     
     
       11. The system according to  claim 10 , wherein the process further comprises:
 receiving, on a user interface of the third device, an indication that the third person selected the selectable option to join the call and, in response, routing the call to the third device of the third person while the call is connected to the second device. 
 
     
     
       12. The system according to  claim 8 , wherein the process further comprises:
 queuing the call by determining a number associated with a first device associated with the first person, wherein the number is used to initiate a second call. 
 
     
     
       13. The system according to  claim 8 , wherein the first score and the second score are determined based on an inflection of a first voice of the first person or of a second voice of the second person. 
     
     
       14. The system according to  claim 8 ,
 wherein the first time is within a pre-determined time window that begins from a start of the conversation on the call, and 
 wherein the second time is outside the pre-determined time window. 
 
     
     
       15. A non-transitory computer-readable medium storing instructions that, when executed by a computing system, cause the computing system to perform operations comprising:
 receiving call data related to a conversation between two or more people including a first person and a second person on a call; 
 determining, using the call data while the first person is on the call with the second person, at least a first score associated with a first time on the call and a second score associated with a second time on the call; 
 determining a third score based on the first score and the second score, wherein the third score indicates a sentiment over time between the first time and the second time; and 
 in response to determining that the third score is less than or equal to a pre-determined score, storing the call in a queue. 
 
     
     
       16. The non-transitory computer-readable medium of  claim 15 , wherein the operations further comprise:
 adjusting the third score based on a time of day when the call occurs, wherein adjusting the third score is performed by a machine learning model trained to adjust sentiment scores, wherein the machine learning model was trained using pairs of words to sentiment values. 
 
     
     
       17. The non-transitory computer-readable medium of  claim 15 , wherein the operations further comprise:
 sending, to a third device associated with a third person, a selectable option to allow the third person to listen or speak on the call while the call is mapped to a second device of the second person. 
 
     
     
       18. The non-transitory computer-readable medium of  claim 17 , wherein the operations further comprise:
 receiving, on a user interface of the third device, an indication that the third person selected the selectable option to join the call and, in response, routing the call to the third device of the third person while the call is connected to the second device. 
 
     
     
       19. The non-transitory computer-readable medium of  claim 15 , wherein the operations further comprise:
 queuing the call by determining a number associated with a first device associated with the first person, wherein the number is used to initiate a second call. 
 
     
     
       20. The non-transitory computer-readable medium of  claim 15 ,
 wherein the first score and the second score are determined based on an inflection of a first voice of the first person or of a second voice of the second person, 
 wherein the first time is within a pre-determined time window that begins from a start of the conversation on the call, and 
 wherein the second time is outside the pre-determined time window.